This is the backtrace of x264. I just decided to run one of the sample programs to test it:
Starting program: /parsec/install/bin/x264 [tcsetpgrp failed in terminal_inferior: Inappropriate ioctl for device] warning: shared library handler failed to enable breakpoint [tcsetpgrp failed in terminal_inferior: Inappropriate ioctl for device] Program received signal SIGSEGV, Segmentation fault. 0x000000012011c688 in ?? () (gdb) backtrace #0 0x000000012011c688 in ?? () #1 0x000000012011ca48 in ?? () #2 0x0000000120100994 in ?? () #3 0x0000000120000218 in ?? () Previous frame inner to this frame (corrupt stack?) I get the same backtrace with or without input data to the test program. -Simran ----- Original Message ----- From: "Ali Saidi" <[email protected]> To: "M5 users mailing list" <[email protected]> Sent: Thursday, November 11, 2010 2:57:48 PM Subject: Re: [m5-users] parsec benchmark on m5 help Slightly de-railing this topic, we probably should add a parsec benchmark to the regression suite. It would catch some errors that we might not otherwise see in terms of FS mp. What are the segfaults. Can you run in GDB and give us a backtrace? Ali On Thu, 11 Nov 2010 14:45:40 -0800 (PST), Simran Basi <[email protected]> wrote: > Well it happens regardless of if i use m5-stable or this changeset. > Do you recommend a different version thats more up to date that > m5-stable that should work? > > -Simran > > ----- Original Message ----- > From: "Gabriel Michael Black" <[email protected]> > To: "M5 users mailing list" <[email protected]>, "ef" > <[email protected]> > Sent: Thursday, November 11, 2010 2:12:32 PM > Subject: Re: [m5-users] parsec benchmark on m5 help > > That's the architecture affected by the most recent change, but not > necessarily the architecture being used. Is the simulator itself > segfaulting, or is it a simulated segfault? The first shouldn't > happen > under any normal circumstances, so if it is we need to fix that. > > Gabe > > Quoting ef <[email protected]>: > >> Parsec doesn't run in SE mode at the moment, and its probably not >> worth it >> to make it do so. >> The parsec images are compiled for ALPHA, it looks like you are >> running x86. >> >> >> On Thu, Nov 11, 2010 at 3:14 PM, Simran Basi <[email protected]> >> wrote: >> >>> Has anyone managed to run the parsec benchmarks on M5 in either >>> full-system >>> or syscall emulation mode? >>> >>> I've also tried running the image provided by this group at UTexas >>> >>> http://www.cs.utexas.edu/~parsec_m5/<http://www.cs.utexas.edu/%7Eparsec_m5/> >>> >>> Has anyone managed to run the benchmarks off this image. When I try >>> to >>> execute the benchmarks they all quit abruptly with a segmentation >>> fault. >>> >>> My current changeset is: >>> >>> changeset: 7715:5581d0cd2bdb >>> tag: tip >>> user: Gabe Black <[email protected]> >>> date: Fri Oct 22 00:24:15 2010 -0700 >>> summary: X86: Make nop a regular, non-microcoded instruction. >>> >>> Please let me know if you've managed to get this to work or have >>> some >>> suggestions on what to look at to fix it >>> >>> Thanks, >>> Simran >>> _______________________________________________ >>> m5-users mailing list >>> [email protected] >>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >>> >> > > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
